Impacts of alternative climate information on hydrologic processes with SWAT: A comparison of NCDC, PRISM and NEXRAD datasets

Abstract

Precipitation and temperature are two primary drivers that significantly affect hydrologic processes in a watershed. A network of land-based National Climatic Data Center (NCDC) weather stations has been typically used as a primary source of climate input for agro-ecosystem models. However, the network may lack the density to adequately capture spatial climate variability throughout large watersheds. High-resolution weather datasets based on 4km×4km grid, such as Next Generation Weather Radar (NEXRAD) and Parameter–Elevation Regressions on Independent Slopes Model (PRISM), have become increasingly available as alternatives to conventional land-based networks. The goal of this study was to evaluate impacts of the three weather datasets, NCDC, NEXRAD, and PRISM, on hydrologic processes in an agricultural catchment in Kansas. A method of collecting and processing three sets of weather input datasets was developed and applied to a calibrated Soil and Water Assessment Tool (SWAT) model for the Smoky Hill River watershed (SHRW) in west-central Kansas, which is sparsely covered by NCDC weather stations with fair to poor range of NEXRAD coverage. SHRW is a typical agricultural catchment in the Central Great Plains; research findings here may be applicable to large areas of the US with similar topography and climate conditions. The SWAT model based on PRISM dataset was able to capture daily streamflow alterations with a greater accuracy compared to NCDC and NEXRAD based SWAT models. With three different weather inputs, SWAT with NCDC consistently overestimated monthly stream discharges, while the SWAT models based on NEXRAD and PRISM datasets tended to underestimate monthly high flows of over 8m3s−1 and overestimate monthly low flows of below 1m3s−1. In general, all models overestimated streamflow in dry years and underestimated streamflow in wet years, however, the PRISM-based model generated smaller bias than the models utilizing NEXRAD or NCDC. The use of PRISM resulted in better statistical performance metrics for streamflow. The conducted study suggests that gridded weather datasets can significantly improve simulated streamflow at daily, monthly and yearly scales as compared to traditional land-based networks.